calculate the left-side noparametric confidence intervals

Currently, I have a set of data in 1-D, and use KDE to predict the distribution. I wander to obatin: 90% left-side confidence interval we created greater than a interger(e.g.,-100)

data set : X --> P(X>=-100)

i suppose it need to use hypothesis testing. the distribution estimation i use Gauss-based KDE, is it possible we use the same as nomoral distribution appraoch to calculate it?

how can i deal with it? thanks

[cindy]

and also I found the code here.

http://www.mathworks.com/matlabcentral/fileexchange/27067-non-parametric-errorconfidence-intervals

but I am not sure it is suitable or not?